[go: up one dir, main page]

US9311293B2 - Techniques for generating translation clusters - Google Patents

Techniques for generating translation clusters Download PDF

Info

Publication number
US9311293B2
US9311293B2 US13/600,301 US201213600301A US9311293B2 US 9311293 B2 US9311293 B2 US 9311293B2 US 201213600301 A US201213600301 A US 201213600301A US 9311293 B2 US9311293 B2 US 9311293B2
Authority
US
United States
Prior art keywords
translation
server
clusters
potential
potential translations
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active, expires
Application number
US13/600,301
Other versions
US20130275118A1 (en
Inventor
John Denero
Mohit Bansal
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Google LLC
Original Assignee
Google LLC
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Google LLC filed Critical Google LLC
Priority to US13/600,301 priority Critical patent/US9311293B2/en
Assigned to GOOGLE INC. reassignment GOOGLE INC. ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: DeNero, John, BANSAL, MOHIT
Publication of US20130275118A1 publication Critical patent/US20130275118A1/en
Application granted granted Critical
Publication of US9311293B2 publication Critical patent/US9311293B2/en
Assigned to GOOGLE LLC reassignment GOOGLE LLC CHANGE OF NAME Assignors: GOOGLE INC.
Active legal-status Critical Current
Adjusted expiration legal-status Critical

Links

Images

Classifications

    • G06F17/2785—
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/30—Semantic analysis
    • G06F17/2827—
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/40—Processing or translation of natural language
    • G06F40/42—Data-driven translation
    • G06F40/45—Example-based machine translation; Alignment

Definitions

  • the present disclosure relates to language translation and, more particularly, to techniques for generating translation clusters.
  • a bilingual lexicon can generally refer to a vocabulary, e.g., a set of words, in two different languages. Specifically, the bilingual lexicon can relate a word in a first language to one or more words in the second language.
  • One type of bilingual lexicon is a bilingual dictionary.
  • a bilingual dictionary can provide meanings or definitions of words in one (a unidirectional bilingual dictionary) or both (a bidirectional bilingual dictionary) of the first and second languages. Bilingual dictionaries, therefore, can also be referred to as translation dictionaries.
  • a unidirectional bilingual dictionary one of the languages may be a source language (spoken/understood by a reader) and the other language may be a target language.
  • a bilingual dictionary may be a Spanish-English dictionary or a Japanese-English dictionary.
  • a computer-implemented technique can include receiving, at a server including one or more processors, a source word in a source language.
  • the technique can include determining, at the server, one or more potential translations for the source word in a target language different than the source language.
  • the technique can include determining, at the server, one or more synonyms for each of the one or more potential translations to obtain a plurality of potential translations.
  • the technique can include determining, at the server, one or more translation clusters using the plurality of potential translations and a first clustering algorithm.
  • Each translation cluster can contain all of the plurality of potential translations that have a similar denotation and each of the plurality of translations that have a similar denotation can be included in a specific translation cluster.
  • Each translation cluster can include at least one distinct potential translation of the plurality of potential translations, and the one or more translation clusters can collectively include all of the plurality of potential translations.
  • the technique can also include outputting, at the server, the one or more translation
  • the first clustering algorithm can be defined as:
  • T S represents the plurality of potential translations
  • C represents a synonym set including a set of target-language words
  • B represents a source-specific synonym set, which is a subset of T S
  • ⁇ represents a set of source-specific synonym sets
  • G represents the one or more translation clusters for T S .
  • determining the one or more synonyms for each the one or more potential translations can include applying a second clustering algorithm over a portion of the target language using one or more corpora.
  • the second clustering algorithm can be a K-means clustering algorithm, K being approximately 1000.
  • the one or more translation clusters are not determined based on context.
  • determining the one or more synonyms for each of the one or more potential translations can include receiving, at the server, the synonyms from a predetermined datastore in response to a request, the predetermined datastore including one or more synonyms for each of a plurality of words in the target language.
  • the technique can further include associating, at the server, a usage example with each of the one or more translation clusters, and outputting, at the server, the one or more usage examples with the one or more translation clusters.
  • the technique can further include determining, at the server, the one or more usage examples for the one or more translation clusters, respectively, by extracting phrase pairs using a parallel corpus.
  • the source word can be input by a user at a computing device and transmitted from the computing device to the server via a network.
  • outputting the one or more translation clusters at the server can include transmitting, from the server, the one or more translation clusters to the computing device via the network for display to the user at the computing device.
  • a server is also presented.
  • the server can include a communication device and one or more processors.
  • the communication device can be configured to receive a source word in a source language.
  • the one or more processors can be configured to determine one or more potential translations for the source word in a target language different than the source language.
  • the one or more processors can be configured to determine one or more synonyms for each of the one or more potential translations to obtain a plurality of potential translations.
  • the one or more processors can also be configured to determine one or more translation clusters using the plurality of potential translations and a first clustering algorithm.
  • Each translation cluster can contain all of the plurality of potential translations that have a similar denotation and each of the plurality of translations that have a similar denotation can be included in a specific translation cluster.
  • Each translation cluster can include at least one distinct potential translation of the plurality of potential translations, and the one or more translation clusters can collectively include all of the plurality of potential translations.
  • the communication device can also
  • the first clustering algorithm can be defined as:
  • T S represents the plurality of potential translations
  • C represents a synonym set including a set of target-language words
  • B represents a source-specific synonym set, which is a subset of T S
  • ⁇ represents a set of source-specific synonym sets
  • G represents the one or more translation clusters for T S .
  • the one or more processors can be configured to determine the one or more synonyms for each the one or more potential translations by applying a second clustering algorithm over a portion of the target language using one or more corpora.
  • the second clustering algorithm can be a K-means clustering algorithm, K being approximately 1000.
  • the one or more translation clusters are not determined based on context.
  • the one or more processors can be configured to determine the one or more synonyms for each of the one or more potential translations by receiving, via the communication device, the synonyms from a predetermined datastore in response to a request, the predetermined datastore including one or more synonyms for each of a plurality of words in the target language.
  • the one or more processors can be further configured to associate a usage example with each of the one or more translation clusters, and the communication device can be further configured to output the one or more usage examples with the one or more translation clusters.
  • the one or more processors can be further configured to determine the one or more usage examples for the one or more translation clusters, respectively, by extracting phrase pairs using a parallel corpus.
  • the source word can be input by a user at a computing device and transmitted from the computing device to the server via a network.
  • the communication device can be configured to output the one or more translation clusters by transmitting the one or more translation clusters to the computing device via the network for display to the user at the computing device.
  • FIG. 1 is a diagram of an example computing network including a server according to some implementations of the present disclosure
  • FIG. 2 is a functional block diagram of the server of FIG. 1 ;
  • FIG. 3 is a flow diagram of an example technique for generating translation clusters according to some implementations of the present disclosure.
  • Bilingual dictionaries are typically created by one or more human lexicographers, which can be both costly and time consuming.
  • machine translation may be based on context of a word or words.
  • techniques are presented for generating translation clusters.
  • the techniques include machine translation and clustering, which can be faster and less expensive than manual generation by one or more human lexicographers. Further, the techniques can automatically generate translation clusters without using context.
  • the techniques generally provide for a more useful and user-friendly presentation of potential translations of a word from a source language to a different target language by arranging the potential translations into clusters.
  • the techniques are further described in the paper attached hereto entitled “Unsupervised Translation Sense Clustering,” which is incorporated by reference in its entirety.
  • the computing network 100 can include a server 104 including one or more processors.
  • the server 104 can be configured according to some implementations of the present disclosure. While one server 104 is shown, it should be appreciated that a plurality of servers can be implemented operating in a parallel or distributed architecture.
  • the computing network 100 can also include a computing device 108 (a desktop computer, a laptop computer, a tablet computer, a mobile phone, etc.).
  • the server 104 can be configured for communication with the computing device 108 via a network 112 .
  • the network 112 can include a local area network (LAN), a wide area network (WAN), e.g., the Internet, or a combination thereof.
  • the computing device 108 can display information to and/or receive input from a user 116 .
  • the user 116 can input a source word in a source language at the computing device 108 .
  • the computing device 108 can transmit the source word to the server 104 via the network 112 .
  • the server 104 can generate one or more translation clusters according to the techniques of the present disclosure (described in detail below). After generating the one or more translation clusters, the server 104 can output the one or more translation clusters. For example, the server 104 can transmit the one or more translation clusters to the computing device 108 via the network 112 , and the computing device 108 can display the one or more translation clusters to the user 116 .
  • the server 104 can be any suitable computing device including a communication device 200 , a processor 204 , a translation control module 208 , a synonym determination module 212 , and a clustering control module 216 . It should be appreciated that while one processor 204 is shown, the use of the term “processor 204 ” herein includes a single processor the server 104 as well as two or more processors operating in a parallel or distributed architecture.
  • the server 104 can also include other suitable components, such as memory 220 .
  • the processor 204 can wholly or partially execute the various modules of the server 104 .
  • the communication device 200 can be configured for communication between the computing device 108 and the server 104 via the network 112 . Specifically, the communication device 200 can receive the source word in the source language from the computing device 108 via the network 112 , and the communication device 200 can transmit the one or more translation clusters to the computing device 108 via the network 112 .
  • the communication device 200 can include any suitable components used for communication via the network 112 , such as a transceiver.
  • the communication device 200 can be configured for communication via one or more communication mediums, such as wireless (e.g., radio frequency or satellite) communication and/or wired (e.g., Ethernet) communication. It should be appreciated, however, that the communication device 200 can be configured for communication via other suitable communication mediums.
  • the processor 204 can communicate with and/or control the communication device 200 , the translation control module 208 , the synonym determination module 212 , the clustering control module 216 , and the memory 220 .
  • the processor 204 can wholly or partially execute the various modules of the computing device 108 .
  • the processor 204 (or processors) can wholly or partially execute the translation control module 208 , the synonym determination module 212 , and the clustering control module 216 .
  • the processor 204 can also control other functions of the computing device 108 including, but not limited to, loading and executing an operating system, controlling communication by the communication device 200 via the network 112 , and controlling various operations relating to the memory 220 (storing, searching, retrieving, deleting, etc.).
  • the memory 220 can be any suitable storage medium capable of storing data, searching the data, and deleting the data.
  • the memory 220 can be non-volatile memory.
  • the translation control module 208 can determine one or more potential translations for the source word. Each of the potential translations can be in the target language, which is different than the source language. For example, the source language may be Spanish and the target language may be English.
  • the translation control module 208 can determine the one or more potential translations using suitable machine translation techniques. For example, the translation control module 208 may use a language model and/or a translation model in obtaining the one or more potential translations from the source word.
  • the synonym determination module 212 can determine one or more synonyms for each of the one or more potential translations to obtain a plurality of potential translations. Rather, the plurality of potential translations can include the one or more potential translations and their various synonyms.
  • the synonym determination module 212 can receive the one or more potential translations from the translation control module 208 (or via the processor 204 ).
  • the synonym determination module 212 can retrieve the synonyms from an external datastore 250 .
  • the external datastore 250 can be accessed via the network 112 using the communication device 200 . While an external datastore 250 is shown and described herein, it should be appreciated that an internal datastore could also be used.
  • the external datastore 250 can include one or more synonyms for each of a plurality of words in the target language.
  • the external datastore 250 can include all of the synonyms for all of the words in the target language.
  • the synonyms can also be divided into sets. These sets of synonyms can also be referred to as “synsets.” For example, each of the synonyms in a set can have a similar denotation.
  • the synonym determination module 212 can determine the synonyms itself.
  • the synonym determination module 212 can determine the synonyms for each of the one or more potential translations by applying a clustering algorithm over a portion of the target language using one or more corpora.
  • the synonym determination module 212 can determine the synonyms based on usage statistics in the one or more corpora.
  • the one or more corpora can include large monolingual corpora and/or large parallel corpora.
  • This clustering algorithm can be distinct from a clustering algorithm used to determine the one or more translation clusters (a first clustering algorithm, described in detail below) and can therefore be referred to as a second clustering algorithm.
  • the second clustering algorithm can be a K-means clustering algorithm.
  • K may be approximately 1000.
  • the selection of K ⁇ 1000 may indicate a minimum value of K that produces a desired level of accuracy without unnecessary additional processing.
  • the clustering control module 216 can determine the one or more translation clusters using the plurality of potential translations and the first clustering algorithm.
  • the clustering control module 216 can receive the plurality of potential translations from the synonym determination module 212 (or via the processor 204 ).
  • Each translation cluster can contain all of the plurality of potential translations that have a similar denotation and each of the plurality of translations that have a similar denotation can be included in a specific translation cluster.
  • Each translation cluster can include at least one distinct potential translation of the plurality of potential translations, and the one or more translation clusters can collectively include all of the plurality of potential translations. In other words, the determination of which translation clusters are the correct translation clusters to obtain the one or more translation clusters can be described as being based on the following properties.
  • the determination does not enumerate all denotations of the source word. Rather, denotation distinctions may be made in situations when they affect cross-lingual lexical choice. For example, if a source words has a plurality of denotations but translates the same way regardless of the denotation, then there may be a single appropriate translation cluster. Second, none of the one or more translation clusters should be a superset of another. This criterion of “completeness” encourages larger translation clusters that are easier to interpret, because their unifying denotations can be identified as the intersection of denotations of the translations in the translation cluster. Lastly, the one or more translation clusters do not need to form a partition of the source word translations. Rather, a potential translation can appear in more than one of the one or more translation clusters.
  • the first clustering algorithm can be defined as follows:
  • T S represents the plurality of potential translations
  • C represents a synset including a set of target-language words
  • B represents a source-specific synset, which is a subset of T S
  • ⁇ represents a set of source-specific synsets
  • G represents the one or more translation clusters for T S . It should be appreciated, however, that other suitable clustering algorithms can be used.
  • the server 104 can associate a usage example with each translation cluster.
  • the usage examples can be predetermined and can be retrieved by the server 104 , e.g., from the external datastore 250 .
  • the server 104 can generate the usage examples itself.
  • the server 104 can determine the usage examples by extracting phrase pairs using a parallel corpus.
  • the server 104 can then output the one or more translation clusters (and, if applicable, the one or more usage examples).
  • the server 104 can transmit this information to the computing device 108 via the network 112 , and the computing device 108 can display the information to the user 116 . In some implementations, this process could be repeated for some or all the words of the source language to automatically generate a bilingual dictionary or a bilingual thesaurus.
  • the source word “colocar” in Spanish can be input to the server 104 .
  • the server 104 can generate the following translation clusters and usage examples for the source word “colocar” in the English language:
  • this layout of the translation clusters, the denotation descriptions, and the usage examples could be displayed to the user 116 at the computing device 108 .
  • the technique 300 begins at 304 .
  • the server 104 can receive a source word in a source language.
  • the server 104 can determine one or more potential translations for the source word in a target language different than the source language.
  • the server 104 can determine one or more synonyms for each of the one or more potential translations to obtain a plurality of potential translations.
  • the server 104 can determine one or more translation clusters using the plurality of potential translations and a first clustering algorithm.
  • Each translation cluster can contain all of the plurality of potential translations that have a similar denotation and each of the plurality of translations that have a similar denotation can be included in a specific translation cluster.
  • Each translation cluster can include at least one distinct potential translation of the plurality of potential translations, and the one or more translation clusters can collectively include all of the plurality of potential translations.
  • the server 104 can output the one or more translation clusters.
  • the technique 300 can then end or return to 304 for one or more additional cycles.
  • Example embodiments are provided so that this disclosure will be thorough, and will fully convey the scope to those who are skilled in the art. Numerous specific details are set forth such as examples of specific components, devices, and methods, to provide a thorough understanding of embodiments of the present disclosure. It will be apparent to those skilled in the art that specific details need not be employed, that example embodiments may be embodied in many different forms and that neither should be construed to limit the scope of the disclosure. In some example embodiments, well-known procedures, well-known device structures, and well-known technologies are not described in detail.
  • first, second, third, etc. may be used herein to describe various elements, components, regions, layers and/or sections, these elements, components, regions, layers and/or sections should not be limited by these terms. These terms may be only used to distinguish one element, component, region, layer or section from another region, layer or section. Terms such as “first,” “second,” and other numerical terms when used herein do not imply a sequence or order unless clearly indicated by the context. Thus, a first element, component, region, layer or section discussed below could be termed a second element, component, region, layer or section without departing from the teachings of the example embodiments.
  • module may refer to, be part of, or include: an Application Specific Integrated Circuit (ASIC); an electronic circuit; a combinational logic circuit; a field programmable gate array (FPGA); a processor or a distributed network of processors (shared, dedicated, or grouped) and storage in networked clusters or datacenters that executes code or a process; other suitable components that provide the described functionality; or a combination of some or all of the above, such as in a system-on-chip.
  • the term module may also include memory (shared, dedicated, or grouped) that stores code executed by the one or more processors.
  • code may include software, firmware, byte-code and/or microcode, and may refer to programs, routines, functions, classes, and/or objects.
  • shared means that some or all code from multiple modules may be executed using a single (shared) processor. In addition, some or all code from multiple modules may be stored by a single (shared) memory.
  • group means that some or all code from a single module may be executed using a group of processors. In addition, some or all code from a single module may be stored using a group of memories.
  • the techniques described herein may be implemented by one or more computer programs executed by one or more processors.
  • the computer programs include processor-executable instructions that are stored on a non-transitory tangible computer readable medium.
  • the computer programs may also include stored data.
  • Non-limiting examples of the non-transitory tangible computer readable medium are nonvolatile memory, magnetic storage, and optical storage.
  • the present disclosure also relates to an apparatus for performing the operations herein.
  • This apparatus may be specially constructed for the required purposes, or it may comprise a general-purpose computer selectively activated or reconfigured by a computer program stored on a computer readable medium that can be accessed by the computer.
  • a computer program may be stored in a tangible computer readable storage medium, such as, but is not limited to, any type of disk including floppy disks, optical disks, CD-ROMs, magnetic-optical disks, read-only memories (ROMs), random access memories (RAMs), EPROMs, EEPROMs, magnetic or optical cards, application specific integrated circuits (ASICs), or any type of media suitable for storing electronic instructions, and each coupled to a computer system bus.
  • the computers referred to in the specification may include a single processor or may be architectures employing multiple processor designs for increased computing capability.
  • the present disclosure is well suited to a wide variety of computer network systems over numerous topologies.
  • the configuration and management of large networks comprise storage devices and computers that are communicatively coupled to dissimilar computers and storage devices over a network, such as the Internet.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Machine Translation (AREA)

Abstract

A computer-implemented technique can include receiving, at a server including one or more processors, a source word in a source language. The technique can include determining, at the server, one or more potential translations for the source word in a target language different than the source language. The technique can include determining, at the server, one or more synonyms for each of the one or more potential translations to obtain a plurality of potential translations. The technique can include determining, at the server, one or more translation clusters using the plurality of potential translations and a clustering algorithm. Each translation cluster can contain all of the plurality of potential translations that have a similar denotation and each of the plurality of translations that have a similar denotation can be included in a specific translation cluster. The technique can also include outputting, at the server, the one or more translation clusters.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS
This application claims the benefit of U.S. Provisional Application No. 61/623,743, filed on Apr. 13, 2012. The entire disclosure of the above application is incorporated herein by reference.
FIELD
The present disclosure relates to language translation and, more particularly, to techniques for generating translation clusters.
BACKGROUND
The background description provided herein is for the purpose of generally presenting the context of the disclosure. Work of the presently named inventors, to the extent it is described in this background section, as well as aspects of the description that may not otherwise qualify as prior art at the time of filing, are neither expressly nor impliedly admitted as prior art against the present disclosure.
A bilingual lexicon can generally refer to a vocabulary, e.g., a set of words, in two different languages. Specifically, the bilingual lexicon can relate a word in a first language to one or more words in the second language. One type of bilingual lexicon is a bilingual dictionary. A bilingual dictionary can provide meanings or definitions of words in one (a unidirectional bilingual dictionary) or both (a bidirectional bilingual dictionary) of the first and second languages. Bilingual dictionaries, therefore, can also be referred to as translation dictionaries. In the case of a unidirectional bilingual dictionary, one of the languages may be a source language (spoken/understood by a reader) and the other language may be a target language. For example only, a bilingual dictionary may be a Spanish-English dictionary or a Japanese-English dictionary.
SUMMARY
A computer-implemented technique is presented. The technique can include receiving, at a server including one or more processors, a source word in a source language. The technique can include determining, at the server, one or more potential translations for the source word in a target language different than the source language. The technique can include determining, at the server, one or more synonyms for each of the one or more potential translations to obtain a plurality of potential translations. The technique can include determining, at the server, one or more translation clusters using the plurality of potential translations and a first clustering algorithm. Each translation cluster can contain all of the plurality of potential translations that have a similar denotation and each of the plurality of translations that have a similar denotation can be included in a specific translation cluster. Each translation cluster can include at least one distinct potential translation of the plurality of potential translations, and the one or more translation clusters can collectively include all of the plurality of potential translations. The technique can also include outputting, at the server, the one or more translation clusters.
In some embodiments, the first clustering algorithm can be defined as:
Figure US09311293-20160412-P00001
 ← {C ∩ Ts : C ∈ ∪t∈T s  
Figure US09311293-20160412-P00002
 t}
Figure US09311293-20160412-P00003
  ← ∅
for  
Figure US09311293-20160412-P00001
  ∈  
Figure US09311293-20160412-P00001
  do
if 
Figure US09311293-20160412-P00004
 
Figure US09311293-20160412-P00001
 ′ ∈ 
Figure US09311293-20160412-P00001
  such that 
Figure US09311293-20160412-P00001
 ⊂ 
Figure US09311293-20160412-P00001
 ′ then
add 
Figure US09311293-20160412-P00001
 to 
Figure US09311293-20160412-P00003
return 
Figure US09311293-20160412-P00003

where TS represents the plurality of potential translations, C represents a synonym set including a set of target-language words,
Figure US09311293-20160412-P00005
represents a set of synonym sets in which a specific potential translation t appears, B represents a source-specific synonym set, which is a subset of TS, β represents a set of source-specific synonym sets, and G represents the one or more translation clusters for TS.
In other embodiments, determining the one or more synonyms for each the one or more potential translations can include applying a second clustering algorithm over a portion of the target language using one or more corpora.
In some embodiments, the second clustering algorithm can be a K-means clustering algorithm, K being approximately 1000.
In other embodiments, the one or more translation clusters are not determined based on context.
In some embodiments, determining the one or more synonyms for each of the one or more potential translations can include receiving, at the server, the synonyms from a predetermined datastore in response to a request, the predetermined datastore including one or more synonyms for each of a plurality of words in the target language.
In other embodiments, the technique can further include associating, at the server, a usage example with each of the one or more translation clusters, and outputting, at the server, the one or more usage examples with the one or more translation clusters.
In some embodiments, the technique can further include determining, at the server, the one or more usage examples for the one or more translation clusters, respectively, by extracting phrase pairs using a parallel corpus.
In other embodiments, the source word can be input by a user at a computing device and transmitted from the computing device to the server via a network.
In some embodiments, outputting the one or more translation clusters at the server can include transmitting, from the server, the one or more translation clusters to the computing device via the network for display to the user at the computing device.
A server is also presented. The server can include a communication device and one or more processors. The communication device can be configured to receive a source word in a source language. The one or more processors can be configured to determine one or more potential translations for the source word in a target language different than the source language. The one or more processors can be configured to determine one or more synonyms for each of the one or more potential translations to obtain a plurality of potential translations. The one or more processors can also be configured to determine one or more translation clusters using the plurality of potential translations and a first clustering algorithm. Each translation cluster can contain all of the plurality of potential translations that have a similar denotation and each of the plurality of translations that have a similar denotation can be included in a specific translation cluster. Each translation cluster can include at least one distinct potential translation of the plurality of potential translations, and the one or more translation clusters can collectively include all of the plurality of potential translations. The communication device can also be configured to output the one or more translation clusters.
In some embodiments, the first clustering algorithm can be defined as:
Figure US09311293-20160412-P00006
 ← {C ∩ Ts : C ∈ ∪t∈T s  
Figure US09311293-20160412-P00007
 t}
Figure US09311293-20160412-P00003
  ← ∅
for 
Figure US09311293-20160412-P00006
 ∈ 
Figure US09311293-20160412-P00006
 do
if 
Figure US09311293-20160412-P00008
 
Figure US09311293-20160412-P00006
 ′ ∈ 
Figure US09311293-20160412-P00006
 such that 
Figure US09311293-20160412-P00006
 ⊂ 
Figure US09311293-20160412-P00006
 ′ then
add 
Figure US09311293-20160412-P00006
 to 
Figure US09311293-20160412-P00003
return 
Figure US09311293-20160412-P00003

where TS represents the plurality of potential translations, C represents a synonym set including a set of target-language words,
Figure US09311293-20160412-P00005
represents a set of synonym sets in which a specific potential translation t appears, B represents a source-specific synonym set, which is a subset of TS, β represents a set of source-specific synonym sets, and G represents the one or more translation clusters for TS.
In other embodiments, the one or more processors can be configured to determine the one or more synonyms for each the one or more potential translations by applying a second clustering algorithm over a portion of the target language using one or more corpora.
In some embodiments, the second clustering algorithm can be a K-means clustering algorithm, K being approximately 1000.
In other embodiments, the one or more translation clusters are not determined based on context.
In some embodiments, the one or more processors can be configured to determine the one or more synonyms for each of the one or more potential translations by receiving, via the communication device, the synonyms from a predetermined datastore in response to a request, the predetermined datastore including one or more synonyms for each of a plurality of words in the target language.
In other embodiments, the one or more processors can be further configured to associate a usage example with each of the one or more translation clusters, and the communication device can be further configured to output the one or more usage examples with the one or more translation clusters.
In some embodiments, the one or more processors can be further configured to determine the one or more usage examples for the one or more translation clusters, respectively, by extracting phrase pairs using a parallel corpus.
In other embodiments, the source word can be input by a user at a computing device and transmitted from the computing device to the server via a network.
In some embodiments, the communication device can be configured to output the one or more translation clusters by transmitting the one or more translation clusters to the computing device via the network for display to the user at the computing device.
Further areas of applicability of the present disclosure will become apparent from the detailed description provided hereinafter. It should be understood that the detailed description and specific examples are intended for purposes of illustration only and are not intended to limit the scope of the disclosure.
BRIEF DESCRIPTION OF THE DRAWINGS
The present disclosure will become more fully understood from the detailed description and the accompanying drawings, wherein:
FIG. 1 is a diagram of an example computing network including a server according to some implementations of the present disclosure;
FIG. 2 is a functional block diagram of the server of FIG. 1; and
FIG. 3 is a flow diagram of an example technique for generating translation clusters according to some implementations of the present disclosure.
 DETAILED DESCRIPTION
Bilingual dictionaries are typically created by one or more human lexicographers, which can be both costly and time consuming. In addition, machine translation may be based on context of a word or words. Accordingly, techniques are presented for generating translation clusters. The techniques include machine translation and clustering, which can be faster and less expensive than manual generation by one or more human lexicographers. Further, the techniques can automatically generate translation clusters without using context. The techniques generally provide for a more useful and user-friendly presentation of potential translations of a word from a source language to a different target language by arranging the potential translations into clusters. The techniques are further described in the paper attached hereto entitled “Unsupervised Translation Sense Clustering,” which is incorporated by reference in its entirety.
Referring now to FIG. 1, an example computing network 100 is illustrated. The computing network 100 can include a server 104 including one or more processors. The server 104 can be configured according to some implementations of the present disclosure. While one server 104 is shown, it should be appreciated that a plurality of servers can be implemented operating in a parallel or distributed architecture. The computing network 100 can also include a computing device 108 (a desktop computer, a laptop computer, a tablet computer, a mobile phone, etc.). The server 104 can be configured for communication with the computing device 108 via a network 112. The network 112 can include a local area network (LAN), a wide area network (WAN), e.g., the Internet, or a combination thereof. The computing device 108 can display information to and/or receive input from a user 116.
Specifically, the user 116 can input a source word in a source language at the computing device 108. In response to a request, e.g., from the user 116, to translate the source word to a target language that is different than the source language, the computing device 108 can transmit the source word to the server 104 via the network 112. In response to receiving the source word, the server 104 can generate one or more translation clusters according to the techniques of the present disclosure (described in detail below). After generating the one or more translation clusters, the server 104 can output the one or more translation clusters. For example, the server 104 can transmit the one or more translation clusters to the computing device 108 via the network 112, and the computing device 108 can display the one or more translation clusters to the user 116.
Referring now to FIG. 2, a functional block diagram of the server 104 is illustrated. The server 104 can be any suitable computing device including a communication device 200, a processor 204, a translation control module 208, a synonym determination module 212, and a clustering control module 216. It should be appreciated that while one processor 204 is shown, the use of the term “processor 204” herein includes a single processor the server 104 as well as two or more processors operating in a parallel or distributed architecture. The server 104 can also include other suitable components, such as memory 220. The processor 204 can wholly or partially execute the various modules of the server 104.
The communication device 200 can be configured for communication between the computing device 108 and the server 104 via the network 112. Specifically, the communication device 200 can receive the source word in the source language from the computing device 108 via the network 112, and the communication device 200 can transmit the one or more translation clusters to the computing device 108 via the network 112. The communication device 200, therefore, can include any suitable components used for communication via the network 112, such as a transceiver. The communication device 200 can be configured for communication via one or more communication mediums, such as wireless (e.g., radio frequency or satellite) communication and/or wired (e.g., Ethernet) communication. It should be appreciated, however, that the communication device 200 can be configured for communication via other suitable communication mediums.
The processor 204 can communicate with and/or control the communication device 200, the translation control module 208, the synonym determination module 212, the clustering control module 216, and the memory 220. As previously described, the processor 204 can wholly or partially execute the various modules of the computing device 108. Specifically, the processor 204 (or processors) can wholly or partially execute the translation control module 208, the synonym determination module 212, and the clustering control module 216. The processor 204 can also control other functions of the computing device 108 including, but not limited to, loading and executing an operating system, controlling communication by the communication device 200 via the network 112, and controlling various operations relating to the memory 220 (storing, searching, retrieving, deleting, etc.). The memory 220 can be any suitable storage medium capable of storing data, searching the data, and deleting the data. For example only, the memory 220 can be non-volatile memory.
The translation control module 208 can determine one or more potential translations for the source word. Each of the potential translations can be in the target language, which is different than the source language. For example, the source language may be Spanish and the target language may be English. The translation control module 208 can determine the one or more potential translations using suitable machine translation techniques. For example, the translation control module 208 may use a language model and/or a translation model in obtaining the one or more potential translations from the source word.
The synonym determination module 212 can determine one or more synonyms for each of the one or more potential translations to obtain a plurality of potential translations. Rather, the plurality of potential translations can include the one or more potential translations and their various synonyms. The synonym determination module 212 can receive the one or more potential translations from the translation control module 208 (or via the processor 204). The synonym determination module 212 can retrieve the synonyms from an external datastore 250. For example, the external datastore 250 can be accessed via the network 112 using the communication device 200. While an external datastore 250 is shown and described herein, it should be appreciated that an internal datastore could also be used. The external datastore 250 can include one or more synonyms for each of a plurality of words in the target language. For example, the external datastore 250 can include all of the synonyms for all of the words in the target language. The synonyms can also be divided into sets. These sets of synonyms can also be referred to as “synsets.” For example, each of the synonyms in a set can have a similar denotation.
Alternatively, the synonym determination module 212 can determine the synonyms itself. The synonym determination module 212 can determine the synonyms for each of the one or more potential translations by applying a clustering algorithm over a portion of the target language using one or more corpora. Specifically, the synonym determination module 212 can determine the synonyms based on usage statistics in the one or more corpora. For example, the one or more corpora can include large monolingual corpora and/or large parallel corpora. This clustering algorithm can be distinct from a clustering algorithm used to determine the one or more translation clusters (a first clustering algorithm, described in detail below) and can therefore be referred to as a second clustering algorithm. The second clustering algorithm can be a K-means clustering algorithm. It should be appreciated, however, that other suitable clustering algorithms can be used. For example, K may be approximately 1000. The selection of K≈1000 may indicate a minimum value of K that produces a desired level of accuracy without unnecessary additional processing. In other words, a value of K=2000 may produce similar accuracy as a value of K=1000, while being more computationally intensive.
The clustering control module 216 can determine the one or more translation clusters using the plurality of potential translations and the first clustering algorithm. The clustering control module 216 can receive the plurality of potential translations from the synonym determination module 212 (or via the processor 204). Each translation cluster can contain all of the plurality of potential translations that have a similar denotation and each of the plurality of translations that have a similar denotation can be included in a specific translation cluster. Each translation cluster can include at least one distinct potential translation of the plurality of potential translations, and the one or more translation clusters can collectively include all of the plurality of potential translations. In other words, the determination of which translation clusters are the correct translation clusters to obtain the one or more translation clusters can be described as being based on the following properties.
First, the determination does not enumerate all denotations of the source word. Rather, denotation distinctions may be made in situations when they affect cross-lingual lexical choice. For example, if a source words has a plurality of denotations but translates the same way regardless of the denotation, then there may be a single appropriate translation cluster. Second, none of the one or more translation clusters should be a superset of another. This criterion of “completeness” encourages larger translation clusters that are easier to interpret, because their unifying denotations can be identified as the intersection of denotations of the translations in the translation cluster. Lastly, the one or more translation clusters do not need to form a partition of the source word translations. Rather, a potential translation can appear in more than one of the one or more translation clusters.
The first clustering algorithm can be defined as follows:
Figure US09311293-20160412-P00009
 ← {C ∩ Ts : C ∈ ∪t∈T s  
Figure US09311293-20160412-P00010
 t}
Figure US09311293-20160412-P00003
 ← ∅
for 
Figure US09311293-20160412-P00009
 ∈ 
Figure US09311293-20160412-P00009
 do
if 
Figure US09311293-20160412-P00011
 
Figure US09311293-20160412-P00009
 ′ ∈ 
Figure US09311293-20160412-P00009
 such that 
Figure US09311293-20160412-P00009
 ⊂ 
Figure US09311293-20160412-P00009
 ′ then
add 
Figure US09311293-20160412-P00009
 to 
Figure US09311293-20160412-P00003
return 
Figure US09311293-20160412-P00003

where TS represents the plurality of potential translations, C represents a synset including a set of target-language words,
Figure US09311293-20160412-P00005
represents a set of synsets in which a specific potential translation t appears, B represents a source-specific synset, which is a subset of TS, β represents a set of source-specific synsets, and G represents the one or more translation clusters for TS. It should be appreciated, however, that other suitable clustering algorithms can be used.
After determining the one or more translation clusters, the server 104 can associate a usage example with each translation cluster. The usage examples can be predetermined and can be retrieved by the server 104, e.g., from the external datastore 250. Alternatively, the server 104 can generate the usage examples itself. For example, the server 104 can determine the usage examples by extracting phrase pairs using a parallel corpus. The server 104 can then output the one or more translation clusters (and, if applicable, the one or more usage examples). For example, the server 104 can transmit this information to the computing device 108 via the network 112, and the computing device 108 can display the information to the user 116. In some implementations, this process could be repeated for some or all the words of the source language to automatically generate a bilingual dictionary or a bilingual thesaurus.
For example only, the source word “colocar” in Spanish can be input to the server 104. The server 104 can generate the following translation clusters and usage examples for the source word “colocar” in the English language:
Translation
Cluster Denotation Description Usage Example
collocate group or chunk together colocar juntas todas los libros
in a certain order or collocate all the books
place side by side
invest, place, make an investment capitales para colocar
put capital to invest
locate, place assign a location to colocar el número de serie
locate the serial number
place, position, put into a certain place colocar en un lugar
put or abstract location put in a place

For example, this layout of the translation clusters, the denotation descriptions, and the usage examples could be displayed to the user 116 at the computing device 108.
Referring now to FIG. 3, an example technique 300 for generating translation clusters is illustrated. The technique 300 begins at 304. At 304, the server 104 can receive a source word in a source language. At 308, the server 104 can determine one or more potential translations for the source word in a target language different than the source language. At 312, the server 104 can determine one or more synonyms for each of the one or more potential translations to obtain a plurality of potential translations. At 316, the server 104 can determine one or more translation clusters using the plurality of potential translations and a first clustering algorithm. Each translation cluster can contain all of the plurality of potential translations that have a similar denotation and each of the plurality of translations that have a similar denotation can be included in a specific translation cluster. Each translation cluster can include at least one distinct potential translation of the plurality of potential translations, and the one or more translation clusters can collectively include all of the plurality of potential translations. At 320, the server 104 can output the one or more translation clusters.
The technique 300 can then end or return to 304 for one or more additional cycles.
Example embodiments are provided so that this disclosure will be thorough, and will fully convey the scope to those who are skilled in the art. Numerous specific details are set forth such as examples of specific components, devices, and methods, to provide a thorough understanding of embodiments of the present disclosure. It will be apparent to those skilled in the art that specific details need not be employed, that example embodiments may be embodied in many different forms and that neither should be construed to limit the scope of the disclosure. In some example embodiments, well-known procedures, well-known device structures, and well-known technologies are not described in detail.
The terminology used herein is for the purpose of describing particular example embodiments only and is not intended to be limiting. As used herein, the singular forms “a,” “an,” and “the” may be intended to include the plural forms as well, unless the context clearly indicates otherwise. The term “and/or” includes any and all combinations of one or more of the associated listed items. The terms “comprises,” “comprising,” “including,” and “having,” are inclusive and therefore specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. The method steps, processes, and operations described herein are not to be construed as necessarily requiring their performance in the particular order discussed or illustrated, unless specifically identified as an order of performance. It is also to be understood that additional or alternative steps may be employed.
Although the terms first, second, third, etc. may be used herein to describe various elements, components, regions, layers and/or sections, these elements, components, regions, layers and/or sections should not be limited by these terms. These terms may be only used to distinguish one element, component, region, layer or section from another region, layer or section. Terms such as “first,” “second,” and other numerical terms when used herein do not imply a sequence or order unless clearly indicated by the context. Thus, a first element, component, region, layer or section discussed below could be termed a second element, component, region, layer or section without departing from the teachings of the example embodiments.
As used herein, the term module may refer to, be part of, or include: an Application Specific Integrated Circuit (ASIC); an electronic circuit; a combinational logic circuit; a field programmable gate array (FPGA); a processor or a distributed network of processors (shared, dedicated, or grouped) and storage in networked clusters or datacenters that executes code or a process; other suitable components that provide the described functionality; or a combination of some or all of the above, such as in a system-on-chip. The term module may also include memory (shared, dedicated, or grouped) that stores code executed by the one or more processors.
The term code, as used above, may include software, firmware, byte-code and/or microcode, and may refer to programs, routines, functions, classes, and/or objects. The term shared, as used above, means that some or all code from multiple modules may be executed using a single (shared) processor. In addition, some or all code from multiple modules may be stored by a single (shared) memory. The term group, as used above, means that some or all code from a single module may be executed using a group of processors. In addition, some or all code from a single module may be stored using a group of memories.
The techniques described herein may be implemented by one or more computer programs executed by one or more processors. The computer programs include processor-executable instructions that are stored on a non-transitory tangible computer readable medium. The computer programs may also include stored data. Non-limiting examples of the non-transitory tangible computer readable medium are nonvolatile memory, magnetic storage, and optical storage.
Some portions of the above description present the techniques described herein in terms of algorithms and symbolic representations of operations on information. These algorithmic descriptions and representations are the means used by those skilled in the data processing arts to most effectively convey the substance of their work to others skilled in the art. These operations, while described functionally or logically, are understood to be implemented by computer programs. Furthermore, it has also proven convenient at times to refer to these arrangements of operations as modules or by functional names, without loss of generality.
Unless specifically stated otherwise as apparent from the above discussion, it is appreciated that throughout the description, discussions utilizing terms such as “processing” or “computing” or “calculating” or “determining” or “displaying” or the like, refer to the action and processes of a computer system, or similar electronic computing device, that manipulates and transforms data represented as physical (electronic) quantities within the computer system memories or registers or other such information storage, transmission or display devices.
Certain aspects of the described techniques include process steps and instructions described herein in the form of an algorithm. It should be noted that the described process steps and instructions could be embodied in software, firmware or hardware, and when embodied in software, could be downloaded to reside on and be operated from different platforms used by real time network operating systems.
The present disclosure also relates to an apparatus for performing the operations herein. This apparatus may be specially constructed for the required purposes, or it may comprise a general-purpose computer selectively activated or reconfigured by a computer program stored on a computer readable medium that can be accessed by the computer. Such a computer program may be stored in a tangible computer readable storage medium, such as, but is not limited to, any type of disk including floppy disks, optical disks, CD-ROMs, magnetic-optical disks, read-only memories (ROMs), random access memories (RAMs), EPROMs, EEPROMs, magnetic or optical cards, application specific integrated circuits (ASICs), or any type of media suitable for storing electronic instructions, and each coupled to a computer system bus. Furthermore, the computers referred to in the specification may include a single processor or may be architectures employing multiple processor designs for increased computing capability.
The algorithms and operations presented herein are not inherently related to any particular computer or other apparatus. Various general-purpose systems may also be used with programs in accordance with the teachings herein, or it may prove convenient to construct more specialized apparatuses to perform the required method steps. The required structure for a variety of these systems will be apparent to those of skill in the art, along with equivalent variations. In addition, the present disclosure is not described with reference to any particular programming language. It is appreciated that a variety of programming languages may be used to implement the teachings of the present disclosure as described herein, and any references to specific languages are provided for disclosure of enablement and best mode of the present invention.
The present disclosure is well suited to a wide variety of computer network systems over numerous topologies. Within this field, the configuration and management of large networks comprise storage devices and computers that are communicatively coupled to dissimilar computers and storage devices over a network, such as the Internet.
The foregoing description of the embodiments has been provided for purposes of illustration and description. It is not intended to be exhaustive or to limit the disclosure. Individual elements or features of a particular embodiment are generally not limited to that particular embodiment, but, where applicable, are interchangeable and can be used in a selected embodiment, even if not specifically shown or described. The same may also be varied in many ways. Such variations are not to be regarded as a departure from the disclosure, and all such modifications are intended to be included within the scope of the disclosure.

Claims (18)

What is claimed is:
1. A computer-implemented method, comprising:
receiving, at a server from a computing device via a network, the server including one or more processors, a single source word in a source language, wherein the single source word is input by a user at the computing device;
determining, at the server, one or more potential translations for the single source word in a target language different than the source language;
determining, at the server, one or more synonyms for each of the one or more potential translations to obtain a plurality of potential translations, wherein the synonyms are stored in a datastore, and wherein the datastore can be accessed via a network;
generating, at the server, one or more translation clusters using the plurality of potential translations and a first clustering algorithm and without using a context of the single source word, each translation cluster containing all of the plurality of potential translations that have a similar denotation each of the plurality of potential translations that have a similar denotation are included in a specific translation cluster, each translation cluster including at least one distinct potential translation of the plurality of potential translations, the one or more translation clusters collectively including all of the plurality of potential translations; and
outputting, from the server to the computing device via the network, information based on the one or more translation clusters,
wherein the first clustering algorithm is defined as:
Figure US09311293-20160412-P00006
 ← {C ∩ Ts : C ∈ ∪t∈T s  
Figure US09311293-20160412-P00007
 t}  ← ∅ for 
Figure US09311293-20160412-P00006
 ∈ 
Figure US09311293-20160412-P00006
 do if 
Figure US09311293-20160412-P00008
 
Figure US09311293-20160412-P00006
 ′ ∈ 
Figure US09311293-20160412-P00006
 such that 
Figure US09311293-20160412-P00006
 ⊂ 
Figure US09311293-20160412-P00006
 ′ then add 
Figure US09311293-20160412-P00006
 to 
Figure US09311293-20160412-P00003
 return 
Figure US09311293-20160412-P00003
where TS represents the plurality of potential translations, C represents a synonym set including a set of target-language words,
Figure US09311293-20160412-P00005
t represents a set of synonym sets in which a specific potential translation t appears, B represents a source-specific synonym set, which is a subset of TS,
Figure US09311293-20160412-P00012
represents a set of source-specific synonym sets, and
Figure US09311293-20160412-P00013
represents the one or more translation clusters for TS.
2. The computer-implemented method of claim 1, wherein determining the one or more synonyms for each the one or more potential translations includes applying a second clustering algorithm over a portion of the target language using one or more corpora.
3. The computer-implemented method of claim 2, wherein the second clustering algorithm is a K-means clustering algorithm, K being approximately 1000.
4. The computer-implemented method of claim 1, wherein determining the one or more synonyms for each of the one or more potential translations includes receiving, at the server, the synonyms from a predetermined datastore in response to a request, the predetermined datastore including one or more synonyms for each of a plurality of words in the target language.
5. The computer-implemented method of claim 1, further comprising associating, at the server, a usage example with each of the one or more translation clusters, wherein the information includes the one or more usage examples.
6. The computer-implemented method of claim 5, further comprising determining, at the server, the one or more usage examples for the one or more translation clusters, respectively, by extracting phrase pairs using a parallel corpus.
7. The computer-implemented method of claim 6, further comprising generating, by the server, a source-target language bilingual dictionary using the one or more translation clusters and the one or more usage examples.
8. The computer-implemented method of claim 7, wherein the information includes the source-target language bilingual dictionary.
9. The computer-implemented method of claim 1, wherein the information includes the one or more translation clusters.
10. A server comprising:
a communication device configured to receive, from a computing device via a network, a single source word in a source language, wherein the single source word is input by a user at the computing device; and
one or more processors configured to:
determine one or more potential translations for the single source word in a target language different than the source language,
determine one or more synonyms for each of the one or more potential translations to obtain a plurality of potential translations, wherein the synonyms are stored in a datastore, and wherein the datastore can be accessed via a network, and
generate one or more translation clusters using the plurality of potential translations and a first clustering algorithm and without using a context of the single source word, each translation cluster containing all of the plurality of potential translations that have a similar denotation each of the plurality of potential translations that have a similar denotation are included in a specific translation cluster, each translation cluster including at least one distinct potential translation of the plurality of potential translations, and the one or more translation clusters collectively including all of the plurality of potential translations,
wherein the communication device is also configured to output, to the computing device via the network, information based on the one or more translation clusters,
wherein the first clustering algorithm is defined as:
Figure US09311293-20160412-P00006
 ← {C ∩ Ts : C ∈ ∪t∈T s  
Figure US09311293-20160412-P00007
 t}  ← ∅ for 
Figure US09311293-20160412-P00006
 ∈ 
Figure US09311293-20160412-P00006
 do if 
Figure US09311293-20160412-P00008
 
Figure US09311293-20160412-P00006
 ′ ∈ 
Figure US09311293-20160412-P00006
 such that 
Figure US09311293-20160412-P00006
 ⊂ 
Figure US09311293-20160412-P00006
 ′ then add 
Figure US09311293-20160412-P00006
 to 
Figure US09311293-20160412-P00003
 return 
Figure US09311293-20160412-P00003
where TS represents the plurality of potential translations, C represents a synonym set including a set of target-language words,
Figure US09311293-20160412-P00005
t represents a set of synonym sets in which a specific potential translation t appears, B represents a source-specific synonym set, which is a subset of TS,
Figure US09311293-20160412-P00012
represents a set of source-specific synonym sets, and
Figure US09311293-20160412-P00013
represents the one or more translation clusters for TS.
11. The server of claim 10, the one or more processors are configured to determine the one or more synonyms for each the one or more potential translations by applying a second clustering algorithm over a portion of the target language using one or more corpora.
12. The server of claim 11, wherein the second clustering algorithm is a K-means clustering algorithm, K being approximately 1000.
13. The server of claim 10, wherein the one or more processors are configured to determine the one or more synonyms for each of the one or more potential translations by receiving, via the communication device, the synonyms from a predetermined datastore in response to a request, the predetermined datastore including one or more synonyms for each of a plurality of words in the target language.
14. The server of claim 10, wherein the one or more processors are further configured to associate a usage example with each of the one or more translation clusters, and wherein the information includes the one or more usage examples.
15. The server of claim 14, wherein the one or more processors are further configured to determine the one or more usage examples for the one or more translation clusters, respectively, by extracting phrase pairs using a parallel corpus.
16. The server of claim 15, wherein the operations further comprise generating a source-target language bilingual dictionary using the one or more translation clusters and the one or more usage examples.
17. The server of claim 16, wherein the information includes the source-target language bilingual dictionary.
18. The server of claim 10, wherein the information includes the one or more translation clusters.
US13/600,301 2012-04-13 2012-08-31 Techniques for generating translation clusters Active 2034-11-02 US9311293B2 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US13/600,301 US9311293B2 (en) 2012-04-13 2012-08-31 Techniques for generating translation clusters

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US201261623743P 2012-04-13 2012-04-13
US13/600,301 US9311293B2 (en) 2012-04-13 2012-08-31 Techniques for generating translation clusters

Publications (2)

Publication Number Publication Date
US20130275118A1 US20130275118A1 (en) 2013-10-17
US9311293B2 true US9311293B2 (en) 2016-04-12

Family

ID=49325873

Family Applications (1)

Application Number Title Priority Date Filing Date
US13/600,301 Active 2034-11-02 US9311293B2 (en) 2012-04-13 2012-08-31 Techniques for generating translation clusters

Country Status (1)

Country Link
US (1) US9311293B2 (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11055496B2 (en) * 2018-08-31 2021-07-06 Samsung Electronics Co., Ltd. Method and apparatus with sentence mapping

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120158398A1 (en) * 2010-12-17 2012-06-21 John Denero Combining Model-Based Aligner Using Dual Decomposition
KR101364774B1 (en) * 2012-12-07 2014-02-20 포항공과대학교 산학협력단 Method for correction error of speech recognition and apparatus
US11443122B2 (en) * 2020-03-03 2022-09-13 Dell Products L.P. Image analysis-based adaptation techniques for localization of content presentation
US11494567B2 (en) * 2020-03-03 2022-11-08 Dell Products L.P. Content adaptation techniques for localization of content presentation
US11455456B2 (en) 2020-03-03 2022-09-27 Dell Products L.P. Content design structure adaptation techniques for localization of content presentation

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080215309A1 (en) * 2007-01-12 2008-09-04 Bbn Technologies Corp. Extraction-Empowered machine translation
US20090259633A1 (en) * 2008-04-15 2009-10-15 Novafora, Inc. Universal Lookup of Video-Related Data
US20110202334A1 (en) * 2001-03-16 2011-08-18 Meaningful Machines, LLC Knowledge System Method and Apparatus
US8983142B1 (en) * 2010-11-10 2015-03-17 Google Inc. Programmatic silhouette attribute determination

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110202334A1 (en) * 2001-03-16 2011-08-18 Meaningful Machines, LLC Knowledge System Method and Apparatus
US20080215309A1 (en) * 2007-01-12 2008-09-04 Bbn Technologies Corp. Extraction-Empowered machine translation
US20090259633A1 (en) * 2008-04-15 2009-10-15 Novafora, Inc. Universal Lookup of Video-Related Data
US8983142B1 (en) * 2010-11-10 2015-03-17 Google Inc. Programmatic silhouette attribute determination

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11055496B2 (en) * 2018-08-31 2021-07-06 Samsung Electronics Co., Ltd. Method and apparatus with sentence mapping

Also Published As

Publication number Publication date
US20130275118A1 (en) 2013-10-17

Similar Documents

Publication Publication Date Title
US9760569B2 (en) Method and system for providing translated result
US11016966B2 (en) Semantic analysis-based query result retrieval for natural language procedural queries
US9311293B2 (en) Techniques for generating translation clusters
US10311146B2 (en) Machine translation method for performing translation between languages
US11055496B2 (en) Method and apparatus with sentence mapping
US20210026846A1 (en) Querying a data graph using natural language queries
US9613029B2 (en) Techniques for transliterating input text from a first character set to a second character set
AU2016269573B2 (en) Input entity identification from natural language text information
US9779087B2 (en) Cross-lingual discriminative learning of sequence models with posterior regularization
US20130185054A1 (en) Techniques for inserting diacritical marks to text input via a user device
US20200233927A1 (en) Context-based translation retrieval via multilingual space
US8825469B1 (en) Techniques for translating documents including tags
Hasler et al. Dynamic topic adaptation for phrase-based mt
Specia et al. Machine translation quality estimation: Applications and future perspectives
US9619457B1 (en) Techniques for automatically identifying salient entities in documents
US10387550B2 (en) Text restructuring
US20140350914A1 (en) Term translation acquisition method and term translation acquisition apparatus
US20160162468A1 (en) Methods and systems for providing universal portability in machine learning
US20180121422A1 (en) Techniques for providing visual translation cards including contextually relevant definitions and examples
US10043511B2 (en) Domain terminology expansion by relevancy
KR20230069872A (en) Apparatus, method and program for for providing foreign language translation and learning services suitable for the user's vocabulary level using a learning webpage
Zhang et al. Topic-based dissimilarity and sensitivity models for translation rule selection
US10386935B2 (en) Input method editor for inputting names of geographic locations
Parkvall et al. Simulating the genesis of Mauritian
Nikoulina et al. Domain adaptation of statistical machine translation models with monolingual data for cross lingual information retrieval

Legal Events

Date Code Title Description
AS Assignment

Owner name: GOOGLE INC., CALIFORNIA

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:DENERO, JOHN;BANSAL, MOHIT;SIGNING DATES FROM 20120731 TO 20120828;REEL/FRAME:028880/0223

STCF Information on status: patent grant

Free format text: PATENTED CASE

AS Assignment

Owner name: GOOGLE LLC, CALIFORNIA

Free format text: CHANGE OF NAME;ASSIGNOR:GOOGLE INC.;REEL/FRAME:044566/0657

Effective date: 20170929

MAFP Maintenance fee payment

Free format text: PAYMENT OF MAINTENANCE FEE, 4TH YEAR, LARGE ENTITY (ORIGINAL EVENT CODE: M1551); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY

Year of fee payment: 4

MAFP Maintenance fee payment

Free format text: PAYMENT OF MAINTENANCE FEE, 8TH YEAR, LARGE ENTITY (ORIGINAL EVENT CODE: M1552); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY

Year of fee payment: 8